Theory of knowledge and justification across the Western philosophical tradition. Covers the tripartite analysis (justified true belief), Gettier problems, foundationalism vs. coherentism vs. reliabilism, skepticism (Cartesian and external world), rationalism vs. empiricism, Kant's synthesis, social and feminist epistemology, pragmatist epistemology, naturalized epistemology, and philosophy of science (Popper, Kuhn, Lakatos, Feyerabend). Use when analyzing claims about knowledge, evidence, certainty, justification, or scientific method.
Epistemology is the study of knowledge — what it is, how we get it, and how we can tell genuine knowledge from mere belief. It asks: What can we know? How can we justify our beliefs? Is certainty possible? These questions are not academic curiosities — they underpin every scientific claim, every legal verdict, every decision made under uncertainty.
Agent affinity: aristotle (logic and epistemology, Opus), beauvoir (existentialism and phenomenology, Opus)
Concept IDs: philo-epistemology-basics, philo-knowledge-justification, philo-empiricism-rationalism
| # | Domain | Core question | Key thinkers |
|---|---|---|---|
| 1 | Tripartite analysis | What is knowledge? | Plato, Gettier |
| 2 | Theories of justification | What makes beliefs justified? | Descartes, BonJour, Goldman |
| 3 | Skepticism | Can we know anything at all? | Descartes, Hume, Putnam |
| 4 |
| Rationalism vs. empiricism |
| Where does knowledge come from? |
| Descartes, Locke, Hume, Leibniz |
| 5 | Kant's synthesis | How do mind and world cooperate? | Kant |
| 6 | Social epistemology | How does knowledge depend on others? | Fricker, Coady, Goldman |
| 7 | Pragmatist epistemology | What role does practice play in knowledge? | James, Dewey, Peirce |
| 8 | Naturalized epistemology | Should epistemology be empirical science? | Quine |
| 9 | Philosophy of science | How does scientific knowledge work? | Popper, Kuhn, Lakatos, Feyerabend |
Classical definition (from Plato's Theaetetus). S knows that P if and only if:
Knowledge = justified true belief (JTB). This analysis dominated epistemology for over two millennia.
Edmund Gettier shattered the JTB analysis with two short counterexamples — three pages that reshaped the field.
Worked example — Gettier Case 1 (simplified):
Smith has strong evidence that Jones will get the job (the company president told him) and that Jones has ten coins in his pocket (Smith counted them). Smith forms the justified belief: "The person who will get the job has ten coins in his pocket."
Unknown to Smith: he himself will get the job, and he himself happens to have ten coins in his pocket.
Smith's belief is:
The Gettier structure: Justified belief + true by coincidence = not knowledge. The subject's justification connects to the wrong truth-maker.
The Gettier problem spawned four decades of epistemological work:
| Response | Strategy | Problem |
|---|---|---|
| No false lemmas (Clark) | Add: justification must not depend on any false belief | Some Gettier cases have no false intermediate step |
| Defeasibility (Lehrer/Paxson) | Add: no true proposition that, if known, would defeat the justification | Defining "defeater" proves circular |
| Causal theory (Goldman 1967) | Knowledge requires an appropriate causal connection between belief and fact | Fails for mathematical and logical knowledge (no causal link) |
| Reliabilism (Goldman 1979) | Justified belief is produced by a reliable cognitive process | Generality problem: which process type? |
| Virtue epistemology (Sosa) | Knowledge is belief that is true because of the knower's intellectual virtue | Spells out "because of" in various ways |
| Knowledge-first (Williamson) | Knowledge is a primitive — don't analyze it in terms of belief + X | Abandons the analytic project |
Core idea. Justified beliefs form a structure like a building. Some beliefs (basic or foundational beliefs) are self-justifying or justified by direct experience. All other beliefs are justified by resting on foundational beliefs through chains of inference.
Classical foundationalism (Descartes). Only beliefs that are certain (indubitable, infallible) count as foundational. Descartes' cogito ("I think, therefore I am") is the paradigm case. The problem: very few beliefs meet this standard, so classical foundationalism threatens to leave most of what we ordinarily claim to know unjustified.
Modest foundationalism. Foundational beliefs need not be certain — they need only have some degree of non-inferential justification. Perceptual beliefs ("I see a red patch") can be foundational even if fallible.
Core idea. Justification is not a one-directional chain from foundations upward but a web of mutually supporting beliefs. A belief is justified to the extent that it coheres with the rest of one's belief system.
The BonJour argument. Laurence BonJour argued that no belief is epistemically basic — even perceptual beliefs depend on background beliefs for their justification ("I can trust my senses in normal conditions" is itself a belief requiring justification).
Worked example — Coherentism in action:
You believe: (a) the thermometer reads 72F, (b) thermometers are generally reliable, (c) the room feels warm, (d) warm rooms are typically around 70-75F. These beliefs mutually support each other. If you discovered (b) was false (the thermometer is broken), the coherence of (a) with the rest of the web would be undermined.
The isolation objection. A coherent set of beliefs could be entirely fictional — the beliefs in a well-constructed novel cohere perfectly but do not constitute knowledge. Coherence alone seems insufficient; some connection to reality is needed.
Core idea (Goldman 1979). A belief is justified if and only if it is produced by a reliable cognitive process — one that produces a high ratio of true beliefs to false beliefs.
Advantage: Explains why perception justifies (perception is reliable) and why wishful thinking does not (it is unreliable), without requiring the subject to have access to the reliability facts.
The generality problem. Every belief-forming process can be described at multiple levels of generality. "Vision" is reliable; "vision in dim light through fog at 200 meters" may not be. Which description determines justification? Reliabilism has not produced a principled answer.
Worked example — Reliabilism on expertise:
A chess grandmaster glances at a board and "sees" that White has a winning position. Is this knowledge? The reliabilist says: the grandmaster's pattern-recognition process, honed over decades, is highly reliable — it produces true assessments at a rate far above chance. So yes, the grandmaster's belief is justified, even if the grandmaster cannot articulate the full reasoning. This captures the intuition that expertise confers epistemic authority.
Descartes (1641) asked: What if an evil demon is systematically deceiving me about everything I experience? If I cannot rule out this possibility, can I know anything about the external world?
The skeptical argument:
This argument uses the closure principle: if S knows P, and S knows that P entails Q, then S knows Q.
| Response | Strategy | Key thinker |
|---|---|---|
| Moorean | "I know I have hands" is more certain than any skeptical premise | G. E. Moore |
| Contextualism | "Know" shifts meaning by context — in ordinary life, we know; in the philosophy room, standards are higher | DeRose, Lewis |
| Relevant alternatives | To know P, I need only rule out relevant alternatives, not every logically possible one | Dretske, Goldman |
| Externalism | Knowledge does not require the ability to refute the skeptic — it requires a reliable connection to truth | Goldman, Nozick |
| Pragmatist | The skeptical hypothesis makes no practical difference — a distinction without a difference is no distinction | James, Peirce |
| Transcendental | Skepticism is self-defeating — it presupposes the very rational standards it claims to undermine | Kant, Stroud |
Worked example — Putnam's brain in a vat (1981):
Hilary Putnam argued that the brain-in-a-vat hypothesis is self-refuting. If I were a brain in a vat, my word "brain" would refer not to actual brains but to whatever the simulation presents as brains. My sentence "I am a brain in a vat" would be false even if I were a brain in a vat — because the words would not mean what they seem to mean. This is an argument from semantic externalism: the meaning of our words depends on our causal connections to the world, not just on what is "in our heads."
| Rationalism | Empiricism | |
|---|---|---|
| Knowledge source | Reason, innate ideas | Sensory experience |
| Paradigm case | Mathematics, logic | Natural science, perception |
| Key thinkers | Descartes, Leibniz, Spinoza | Locke, Berkeley, Hume |
| Innate ideas? | Yes — some knowledge is hardwired | No — the mind begins as blank slate (tabula rasa) |
Descartes argued that the senses are unreliable (the wax argument: a piece of wax looks, smells, and feels completely different when melted, yet I know it is the same wax — this knowledge comes from the intellect, not the senses). He found certainty only in the deliverances of pure reason: the cogito, the idea of God, and clear and distinct ideas.
David Hume (1711-1776) pushed empiricism to its radical conclusions:
The fork. All genuine knowledge falls into two categories:
The problem of induction. We believe the future will resemble the past (the sun will rise, bread will nourish). But what justifies this belief? Not reason (there is no logical contradiction in the sun failing to rise). Not experience (appealing to past experience to justify reliance on experience is circular). Hume concluded that inductive reasoning has no rational justification — it rests on custom and habit.
Worked example — Hume's problem applied to science:
A scientist observes that water boils at 100C at sea level in every experiment she has ever conducted (say 10,000 observations). She concludes: "Water boils at 100C at sea level." Hume asks: What justifies the inference from 10,000 past observations to the universal claim? The only answer seems to be: "Because nature is uniform." But how do we know nature is uniform? Only from past experience of uniformity — which is circular.
This is not merely academic. The problem of induction is unresolved. Popper, Goodman, Carnap, and Bayesians have all attempted solutions, but none is universally accepted.
Immanuel Kant (1724-1804) was "awakened from his dogmatic slumber" by Hume and sought to reconcile rationalism and empiricism.
Transcendental idealism. We do not know things as they are in themselves (noumena). We know things as they appear to us (phenomena), structured by the forms of our intuition (space and time) and the categories of our understanding (causality, substance, etc.).
Synthetic a priori knowledge. Kant's revolution was to argue that some knowledge is both informative about the world (synthetic) and knowable independently of particular experience (a priori). Example: "Every event has a cause." This is not analytic (the concept of "event" does not logically contain "cause"), but it is known prior to experience because causality is a necessary condition for experience itself.
The Copernican revolution in epistemology. Instead of asking how our knowledge conforms to objects, Kant asked how objects must conform to our knowledge. The mind actively structures experience — it does not passively receive it. This is why we can have a priori knowledge: we are discovering the structure we ourselves impose.
Core idea. Knowledge is not a purely individual achievement — it depends on testimony, institutions, social structures, and power relations.
Most of what we know, we know because someone told us. How is testimonial knowledge possible? C. A. J. Coady (1992) argued that testimony is a basic source of knowledge, not reducible to perception or inference. We are entitled to believe what others tell us unless we have specific reasons for doubt.
Miranda Fricker identified two forms:
Worked example — Epistemic injustice in medical settings:
Studies consistently show that women's reports of pain are taken less seriously than men's — they are more likely to be prescribed sedatives (for anxiety) rather than painkillers. This is testimonial injustice: the patient's testimony about her own experience is discounted due to prejudice. The epistemic wrong compounds the practical harm.
Sandra Harding and others argued that social position shapes epistemic access. Marginalized groups may have epistemic advantages regarding certain phenomena — not because oppression confers magical insight, but because navigating dominant and subordinate social worlds gives a broader perspective. A factory worker understands both management's stated reasons and the shop-floor reality; management typically understands only its own perspective.
Peirce's fallibilism. Charles Sanders Peirce (1839-1914) held that all beliefs are fallible — none is beyond revision. Truth is the ideal limit of inquiry: the belief that the community of inquirers would converge upon given unlimited investigation.
James's pragmatic theory of truth. William James (1842-1910) argued that a belief is true insofar as it is useful — "truth is what works." This is frequently misunderstood as crass. James meant that truth is not a static property of propositions but an active relation between beliefs and experience: true beliefs are those that successfully guide action over time.
Dewey's reflective thinking. John Dewey (1859-1952) argued that inquiry begins with a felt difficulty — a problematic situation — and proceeds through hypothesis, testing, and resolution. Knowledge is not spectating on reality but actively transforming problematic situations into resolved ones.
Worked example — Dewey's inquiry applied to debugging software:
Dewey would say this is the structure of ALL knowledge — not just applied problem-solving but science, philosophy, and everyday learning. Epistemology is not about passive contemplation of eternal truths but about the logic of inquiry.
Quine's proposal (1969). W. V. O. Quine argued that traditional epistemology had failed to provide a non-circular justification of empirical knowledge. His solution: replace normative epistemology with descriptive psychology. Study how humans actually form beliefs from sensory input — make epistemology a branch of empirical science.
The normative objection. Many epistemologists responded that Quine's proposal abandons the essential task of epistemology — evaluating whether beliefs are justified, not merely describing how they are formed. A purely descriptive epistemology cannot distinguish knowledge from superstition.
The compromise. Most contemporary epistemologists accept that empirical findings (cognitive science, psychology of reasoning, statistics of human error) are relevant to epistemology without accepting that epistemology is reducible to empirical science. Normative questions remain.
Karl Popper (1902-1994) proposed that science advances not by confirming hypotheses but by attempting to falsify them. A theory is scientific if and only if it is falsifiable — if there are observations that could show it to be false. Theories that cannot be falsified (Freudian psychoanalysis, by Popper's analysis) are not scientific.
The asymmetry of confirmation. No number of white swan observations proves "all swans are white," but one black swan disproves it. Science should seek black swans.
Thomas Kuhn (1922-1996) argued in The Structure of Scientific Revolutions (1962) that science does not progress by steady accumulation but through revolutionary shifts:
The incommensurability thesis. Kuhn argued that competing paradigms are partly incommensurable — scientists on opposite sides of a paradigm shift literally see the world differently. This raised concerns about scientific rationality: if paradigm choice is not purely rational, is science just a sociological phenomenon?
Lakatos's research programmes. Imre Lakatos (1922-1974) tried to salvage scientific rationality from Kuhn. A research programme has a "hard core" of non-negotiable assumptions protected by a "protective belt" of auxiliary hypotheses. Programmes are progressive (making novel predictions that are confirmed) or degenerating (making only ad hoc adjustments). Rational scientists choose progressive programmes.
Feyerabend's epistemological anarchism. Paul Feyerabend (1924-1994) argued that the only universal methodological principle that does not inhibit scientific progress is "anything goes." Historical case studies show that successful scientists routinely violated every proposed methodological rule.
Worked example — Comparing Popper and Kuhn on the same event:
In 1919, Eddington's eclipse expedition confirmed Einstein's prediction of light bending around the sun.
Popper's reading: Einstein's theory made a bold, precise, falsifiable prediction. Eddington's observation could have falsified it but did not. This is the hallmark of good science — risky prediction followed by corroboration.
Kuhn's reading: Eddington's expedition was conducted by scientists already sympathetic to relativity. The data were ambiguous (some plates supported Newton, some Einstein). Eddington chose to emphasize the plates favoring Einstein. The "confirmation" was partly a social act within a community undergoing paradigm shift.
Both readings capture something real. The tension between them defines much of 20th-century philosophy of science.